parity checks

[squirrellydw]

When doing parity checks in version 6 should I have it write corrections to parity disk set to yes or no?

[JonathanM]

When doing parity checks in version 6 should I have it write corrections to parity disk set to yes or no?

That can be argued both ways.

 

If you are asking the question, then the answer is definitely yes. If you research how parity works and why some people choose no, then you can make up your own mind, but without knowing exactly why and in what situations no is appropriate, just choose yes.

[SSD]

When doing parity checks in version 6 should I have it write corrections to parity disk set to yes or no?

 

Normally, the answer is yes. You might ask yourself, before starting a parity check, what would happen if there was a sync error - would I want parity corrected to be aligned with the data on the disks. If the answer is yes, then you want a correcting check.

 

But there are times, dealing with failing disks and reconstructing arrays with trusting parity, that you may be unsure that the array is properly reassembled. And if you ran a parity check and started getting lots of mismatches, you would know you did something wrong and would NOT want to be updating parity due to sync errors. In such a case you would want a non-correcting check.

 

So about 95% of the time the answer is a yes, and 5% of the time it is no. If in doubt, say "no". If there are no parity mismatches (which most of the time there are not), no and yes result in the same outcome. Just know, if you say "no", and you do have a sync error, you will likely need to rerun a correcting check to fix it.